Over 285,000 total hip replacements are performed annually in the US. With several different designs and material combinations available for use, there is minimal evidence to help guide the selection of a particular implant. The small clinical trials and post-marketing studies that have been completed have three significant limitations: 1) they are underpowered to detect rare events (device failure and complications), 2) they are too short to assess device longevity, and 3) they are limited to single sites or regions with highly skilled surgeons which may not be reflective of national trends. To date, large studies using health insurance claims databases have not been feasible because detailed information on the implanted devices has not been available and information from registries lack patient level characteristics crucial to adequately adjust for confounding. There is a critical need for comprehensive data that compare joint replacement outcomes across types of implants to assist physicians and patients in device selection. The overall goal of this dissertation is to estimate the effect of bearing surface material and femoral head size of hip replacements on the rate of revision, using data that overcome the limitations of the previous studies through the linkage of detailed clinical and administrative data from the Veterans Health Administration (VHA). The advantage over existing registries is its size and the capability of linking to other VHA data sources containing patient-level characteristics to aid in confounding adjustment and identification of a range of important health outcomes. The specific aims of this dissertation are: 1) For metal-on-metal total hip replacements, estimate the effect of femoral head size on the rate of revision surgery; 2) For all other total hip replacements, estimate the effect of femoral head size on the rate of revision surgery; 3) To estimate the effect of bearing surface material in total hip replacements on the rate of revision surgery. A series of high-dimensional propensity-matched cohorts will be created to complete the described analysis among patients receiving a primary total hip replacement at a VHA health center. The proposed dissertation is significant because the US population is rapidly aging and hip replacement will affect an increasing number of people. The approach is innovative because it will combine existing data from multiple sources within the VHA system to allow for the study of long-term outcomes of hip replacement with proper adjustment for patient characteristics. The expected outcome of this work is comparative effectiveness information on the rates of revision among different specific devices, which will aid physicians and patients with device selection.